Generate a project's whole set of art from a repeatable script
Four or more companies' docs show a full fit.
What counts as done
Every listed asset produced and saved to the repository in one run, re-runnable with seeds or the same model to get matching style, with the total cost estimable before the run.
Measured, for our own tests: the total cost is within 10 per cent of that estimate. No docs we read state these figures, so they don't count against a listing here.
How agents ask for it: make all the card art for my game
, generate the app icon, banners and avatars
, make an image for every shot in this script
, regenerate the visuals in the same style
Does the job 4
The docs show it meets the test above, and nothing we read says otherwise.
| Listing | Why |
|---|---|
| Bria API BB Bria Artificial Intelligence Ltd. | seed plus the returned structured_prompt recreate a result precisely, per-image prices are fixed, webhooks carry an HMAC signature the SDKs verify (acc-23 was closed as a key mismatch), and results stay 3 days. source |
| FLORA B FLORA AND FAUNA FLOWERSHOP, INC. | The REST API takes seed in params, quote: true gives the exact request's dollar estimate, Idempotency-Key prevents double runs, webhooks are HMAC-signed, and outputs stay on the team's canvas. source |
| Iris-3B C Speridlabs | --txt-file runs one prompt per line, --seed reproduces an image, --outdir sets where files land, and local Apache-2.0 weights never retire or bill, given an NVIDIA GPU. source |
| Recraft API D Recraft Inc. | Fixed per-image prices, a credits field on every response, random_seed and reusable style_id make a pre-run estimate and a matching re-run possible; calls are synchronous, so no webhook is needed, and acc-29 plans precisely this script. source |
Does the job, with a limit 9
It does the job, but something in the docs or a user's account will get in an agent's way. The limit is named in each row.
| Listing | Why |
|---|---|
| Adobe Firefly API D Adobe Inc. | seeds and numVariations make re-runs repeatable and custom models hold a style, but there is no public price to estimate against, no webhook, result URLs expire in an hour, and the default limit is 4 requests a minute. source |
| Black Forest Labs FLUX API B Black Forest Labs Inc. | Submits return cost in credits and FLUX.2 takes seed and a signed webhook, but results expire after 10 minutes, FLUX 3 Image has no seed and only a `latest` version, and the terms allow changing the service without notice. source |
| fal image models BB Features & Labels, Inc. | A pricing estimate endpoint, per-model seeds and CDN retention of at least 7 days fit a batch, but there is no idempotency key so a resubmit bills twice, and many text-to-image endpoints are marked deprecated without dates. source |
| Ideogram API D Ideogram AI, Inc. | seed, num_images, dry_run quotes and Ed25519-signed webhooks fit a batch script, but links expire after an unstated time, there is no changelog, and acc-33 saw 61 operations removed from the live spec in five days. source |
| Leonardo.Ai API F Leonardo Interactive Pty Ltd | seed, quantity, a webhook callback and apiCreditCost on each response fit a batch, but per-image prices aren't published for an estimate, the webhook is a bearer key with no signature, and models have been retired at zero to 28 days' notice. source |
| OpenAI Image API BB OpenAI OpCo, LLC | Batch pricing and usage tokens exist, but there is no seed parameter, the per-image cost depends on tokens the official calculator underestimates (acc-07), and image models get only three months' retirement notice (acc-01). source |
| Replicate image models B Replicate, LLC | Community model versions can be pinned and webhooks are signed with documented verification, but API outputs are deleted after an hour, and no idempotency key or retirement policy for official models was found. source |
| Seedream on BytePlus ModelArk C BytePlus Pte. Ltd. | Sequential generation returns up to 15 related images in one request with per-image errors and per-image prices billed only on success, but no seed parameter was found, URLs expire in 24 hours, and there is no webhook for images. source |
| Stability AI Image API F Stability AI Ltd | seed, fixed credits per successful result and bytes returned in the response make re-runs and estimates exact, but there is no batch or webhook, and SD 1.6 and Stable Video were removed on the day the notice was posted (24 July 2025). source |
Couldn't check 1
The docs were closed to us, or silent. These carry no mark either way.
| Listing | Why |
|---|---|
| Qwen-Image 2.1 (Model Studio) B Alibaba Cloud (Singapore) Private Limited | Model Studio docs (www.alibabacloud.com and docs.modelstudio.console.alibabacloud.com) did not resolve from here, so seed, price, retention and batch behaviour of the API weren't read. |
Doesn't do this job 1
Listings in the category that do something else, with the reason.
| Listing | Why |
|---|---|
| SurgePix API F SurgePix | Skills return one download URL per request with no seed, model choice or per-call price, so a re-run can't be matched or a cost estimated; acc-24 found only 3 of 9 advertised skills install. source |
What stops agents finishing
- model behind the project's style is retired
- cost calculator underestimates the real charge
- webhook signatures fail verification so batch results cannot be trusted
- results deleted within minutes before the script downloads them
Around it
Before: llm.write-script. After: video-generation.code-rendered-video, code.commit-assets.
Where this comes from
Weight 14 of 100 in Image generation: 7 of 37 accounts; Claude Code scripts calling Recraft for a trading card game, agents told to use real image models for a Discord bot's art, 65 stills per video episode, a game style lost to a deprecation
The same as data: /api/v1/use-cases/image-generation.project-asset-batch.json, or the MCP tool get_use_case with id image-generation.project-asset-batch. Other jobs in this category: /api/v1/niches?category=image-generation.
For companies
Do agents find, use and choose your tools?
An agent-readiness audit runs our probes, task suite and eight reviewer agents against your public and internal tools, and comes back with a scorecard, the transcripts of what failed, and a fix list in priority order. From $2,500, re-run included. We never take payment to move a rank. We do help companies earn one.